# Geometric diagram parsing on SGDP-1K (test)

Source: [Geometric diagram parsing on SGDP-1K (test)](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-14T12:21:58.589Z

Evaluation of Geometric Diagram Parsing performance on the test split of the SGDP-1K dataset, reporting precision, recall, and F1 scores for geometric primitives (Points, Lines, Circles, Planes) and an overall score.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Geometric diagram parsing |
| Dataset (including any stated split) | SGDP-1K (test) |
| Metric | Point Precision |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 12 of 12 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Point Precision](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Point%20Precision>): Higher is better; unit: PERCENT.

- [Point Recall](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Point%20Recall>): Higher is better; unit: PERCENT.

- [Point F1](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Point%20F1>): Higher is better; unit: PERCENT.

- [Line Precision](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Line%20Precision>): Higher is better; unit: PERCENT.

- [Line Recall](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Line%20Recall>): Higher is better; unit: PERCENT.

- [Line F1](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Line%20F1>): Higher is better; unit: PERCENT.

- [Circle Precision](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Circle%20Precision>): Higher is better; unit: PERCENT.

- [Circle Recall](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Circle%20Recall>): Higher is better; unit: PERCENT.

- [Circle F1](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Circle%20F1>): Higher is better; unit: PERCENT.

- [Plane Precision](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Plane%20Precision>): Higher is better; unit: PERCENT.

- [Plane Recall](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Plane%20Recall>): Higher is better; unit: PERCENT.

- [Plane F1](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Plane%20F1>): Higher is better; unit: PERCENT.

- [Solid Accuracy](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Solid%20Accuracy>): Higher is better; unit: PERCENT.

- [Overall Score](<https://www.sota2.com/research/sota/geometric-diagram-parsing-on-sgdp-1k-test.md?metric=Overall%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Point Precision | Result date | Report paper |
| --- | --- | --- | --- |
| Gemini-3-Flash | 99.4 | Not recorded | 69ddca7c7d0172140ec79f7f |
| Qwen3-VL-235B-Thinking | 99.2 | Not recorded | 69ddca7c7d0172140ec79f7f |
| GDP-4B-RL | 99.2 | Not recorded | 69ddca7c7d0172140ec79f7f |
| Qwen3-VL-4B-Instruct | 99 | Not recorded | 69ddca7c7d0172140ec79f7f |
| Qwen3-VL-32B-Instruct | 99 | Not recorded | 69ddca7c7d0172140ec79f7f |
| Qwen3-VL-8B-Instruct | 98.9 | Not recorded | 69ddca7c7d0172140ec79f7f |
| GPT-5.2-1211 | 98.9 | Not recorded | 69ddca7c7d0172140ec79f7f |
| GDP-4B-SFT | 98.9 | Not recorded | 69ddca7c7d0172140ec79f7f |
| Qwen3-VL-235B-Instruct | 98.7 | Not recorded | 69ddca7c7d0172140ec79f7f |
| InternVL3.5-8B-Instruct | 98.5 | Not recorded | 69ddca7c7d0172140ec79f7f |
| Claude-4.5-Sonnet | 95.7 | Not recorded | 69ddca7c7d0172140ec79f7f |
| LLaVA-OneVision-1.5-7B | 91.6 | Not recorded | 69ddca7c7d0172140ec79f7f |

### Gemini-3-Flash

Point Precision: 99.4. Result ID: 69de314947f34ce924c6fb4b.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

Recorded evaluation configuration: {"Access":"Closed-source"}. This record does not assert independent reproduction.

### Qwen3-VL-235B-Thinking

Point Precision: 99.2. Result ID: 69de314947f34ce924c6fb47.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Access":"Open-source","Model Configuration":"Thinking"}. This record does not assert independent reproduction.

### GDP-4B-RL

Point Precision: 99.2. Result ID: 69de314947f34ce924c6fb4d.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

Recorded evaluation configuration: {"Training Strategy":"RL"}. This record does not assert independent reproduction.

### Qwen3-VL-4B-Instruct

Point Precision: 99. Result ID: 69de314947f34ce924c6fb43.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Access":"Open-source","Model Configuration":"Instruct"}. This record does not assert independent reproduction.

### Qwen3-VL-32B-Instruct

Point Precision: 99. Result ID: 69de314947f34ce924c6fb46.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Access":"Open-source","Model Configuration":"Instruct"}. This record does not assert independent reproduction.

### Qwen3-VL-8B-Instruct

Point Precision: 98.9. Result ID: 69de314947f34ce924c6fb45.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Access":"Open-source","Model Configuration":"Instruct"}. This record does not assert independent reproduction.

### GPT-5.2-1211

Point Precision: 98.9. Result ID: 69de314947f34ce924c6fb4a.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

Recorded evaluation configuration: {"Access":"Closed-source"}. This record does not assert independent reproduction.

### GDP-4B-SFT

Point Precision: 98.9. Result ID: 69de314947f34ce924c6fb4c.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

Recorded evaluation configuration: {"Training Strategy":"SFT"}. This record does not assert independent reproduction.

### Qwen3-VL-235B-Instruct

Point Precision: 98.7. Result ID: 69de314947f34ce924c6fb48.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Access":"Open-source","Model Configuration":"Instruct"}. This record does not assert independent reproduction.

### InternVL3.5-8B-Instruct

Point Precision: 98.5. Result ID: 69de314947f34ce924c6fb44.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7af01e209131cea85dc9.md>)

Recorded evaluation configuration: {"Access":"Open-source","Model Configuration":"Instruct"}. This record does not assert independent reproduction.

### Claude-4.5-Sonnet

Point Precision: 95.7. Result ID: 69de314947f34ce924c6fb49.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

Recorded evaluation configuration: {"Access":"Closed-source"}. This record does not assert independent reproduction.

### LLaVA-OneVision-1.5-7B

Point Precision: 91.6. Result ID: 69de314947f34ce924c6fb42.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca7c7d0172140ec79f7f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756f1e209131cea67e58.md>)

[Implementation](<https://llava-vl.github.io/blog/llava-onevision>)

Recorded evaluation configuration: {"Access":"Open-source"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
